I present my Illustrious character LoRA for Magister Umbric from World of Warcraft - Midnight expansion.<hr /><hr />V2 is my attempt to make the hair tendrils more consistent. I also got better screenshots of the outfit but it seems the AI really hates that split collar but I'm tired of retraining this model because it's "close enough." The hair at least looks better and he looks better in NSFW images than in V1. You may need to add horns to the negative prompt.Recommended LoRA strength: 0.6-0.7 Also, I use an adetailer in my generations for this LoRA.V2 main tags are as follows:<pre><code>umbric, 1boy, elf, purple skin, short hair, hair spikes, long eyebrows, blue eyes, blue sclera, solid eyes, no pupils</code></pre>V2 outfit tags are as follows:<pre><code>blue-flamed shoulders, white and silver robe, black belt, two-tone fingerless gloves, black footwear</code></pre>V2 optional tags:<pre><code>staff</code></pre>The settings I used for my generation testing were 3-4 CFG, 20 steps, DPM++ 2M SDE - Karras on Forge with an adetailer (face_yolov8s).<hr /><hr />V1 is my initial version and produces decent results. The shoulders like to ignore the flame effects and the robe has some random adornments not seen in-game but it's close enough. His hair also likes to randomize the tentacle spikes that are supposed to be there but it is what it is. I am still impressed it turned out as well as it did, anyway. You may need to add horns to the negative prompt. Recommended LoRA strength: 0.7 Also, I use an adetailer in my generations for this LoRA.V1 main tags are as follows:<pre><code>umbric, 1boy, elf, blue skin, short spiked hair, long eyebrows, blue eyes, solid eyes, no pupils, glowing eyes</code></pre>V1 outfit tags are as follows:<pre><code>blue-flamed shoulders, white robe, black belt, two-tone fingerless gloves, black footwear</code></pre>V1 optional tags:<pre><code>staff</code></pre>The settings I used for my generation testing were 3-4 CFG, 20 steps, DPM++ 2M SDE - Karras on Forge with an adetailer (face_yolov8s).

Files with SHA256: d49da0c2c8c9b636bfc9f959bbc236713fecf37b3c8e0983540adfd86a848294